Upload policy weights, train config and readme
Browse files- README.md +4 -4
- config.json +31 -45
- model.safetensors +2 -2
- train_config.json +45 -64
README.md
CHANGED
|
@@ -2,15 +2,15 @@
|
|
| 2 |
datasets: quincyu/trace_robotwin
|
| 3 |
library_name: lerobot
|
| 4 |
license: apache-2.0
|
| 5 |
-
model_name:
|
| 6 |
pipeline_tag: robotics
|
| 7 |
tags:
|
| 8 |
-
- tg_diffusion
|
| 9 |
-
- lerobot
|
| 10 |
- robotics
|
|
|
|
|
|
|
| 11 |
---
|
| 12 |
|
| 13 |
-
# Model Card for
|
| 14 |
|
| 15 |
<!-- Provide a quick summary of what the model is/does. -->
|
| 16 |
|
|
|
|
| 2 |
datasets: quincyu/trace_robotwin
|
| 3 |
library_name: lerobot
|
| 4 |
license: apache-2.0
|
| 5 |
+
model_name: trace_act
|
| 6 |
pipeline_tag: robotics
|
| 7 |
tags:
|
|
|
|
|
|
|
| 8 |
- robotics
|
| 9 |
+
- trace_act
|
| 10 |
+
- lerobot
|
| 11 |
---
|
| 12 |
|
| 13 |
+
# Model Card for trace_act
|
| 14 |
|
| 15 |
<!-- Provide a quick summary of what the model is/does. -->
|
| 16 |
|
config.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
| 1 |
{
|
| 2 |
-
"type": "
|
| 3 |
-
"n_obs_steps":
|
| 4 |
"input_features": {
|
| 5 |
"observation.state": {
|
| 6 |
"type": "STATE",
|
|
@@ -53,7 +53,7 @@
|
|
| 53 |
]
|
| 54 |
}
|
| 55 |
},
|
| 56 |
-
"device": "
|
| 57 |
"use_amp": false,
|
| 58 |
"use_peft": false,
|
| 59 |
"push_to_hub": true,
|
|
@@ -62,53 +62,39 @@
|
|
| 62 |
"tags": null,
|
| 63 |
"license": null,
|
| 64 |
"pretrained_path": null,
|
| 65 |
-
"
|
| 66 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 67 |
"ee_pixel_image_resolution": 256,
|
| 68 |
"normalization_mapping": {
|
| 69 |
"VISUAL": "MEAN_STD",
|
| 70 |
-
"STATE": "
|
| 71 |
"ACTION": "MIN_MAX"
|
| 72 |
},
|
| 73 |
-
"drop_n_last_frames": 7,
|
| 74 |
"vision_backbone": "resnet18",
|
| 75 |
-
"
|
| 76 |
-
|
| 77 |
-
|
| 78 |
-
|
| 79 |
-
"
|
| 80 |
-
"
|
| 81 |
-
"
|
| 82 |
-
"
|
| 83 |
-
"
|
| 84 |
-
"
|
| 85 |
-
|
| 86 |
-
|
| 87 |
-
|
| 88 |
-
|
| 89 |
-
"
|
| 90 |
-
"
|
| 91 |
-
"
|
| 92 |
-
"
|
| 93 |
-
"noise_scheduler_type": "DDPM",
|
| 94 |
-
"num_train_timesteps": 100,
|
| 95 |
-
"beta_schedule": "squaredcos_cap_v2",
|
| 96 |
-
"beta_start": 0.0001,
|
| 97 |
-
"beta_end": 0.02,
|
| 98 |
-
"prediction_type": "epsilon",
|
| 99 |
-
"clip_sample": true,
|
| 100 |
-
"clip_sample_range": 1.0,
|
| 101 |
-
"num_inference_steps": null,
|
| 102 |
-
"do_mask_loss_for_padding": false,
|
| 103 |
"compile_model": false,
|
| 104 |
-
"compile_mode": "
|
| 105 |
-
"optimizer_lr": 0.0001,
|
| 106 |
-
"optimizer_betas": [
|
| 107 |
-
0.95,
|
| 108 |
-
0.999
|
| 109 |
-
],
|
| 110 |
-
"optimizer_eps": 1e-08,
|
| 111 |
-
"optimizer_weight_decay": 1e-06,
|
| 112 |
-
"scheduler_name": "cosine",
|
| 113 |
-
"scheduler_warmup_steps": 500
|
| 114 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"type": "trace_act",
|
| 3 |
+
"n_obs_steps": 1,
|
| 4 |
"input_features": {
|
| 5 |
"observation.state": {
|
| 6 |
"type": "STATE",
|
|
|
|
| 53 |
]
|
| 54 |
}
|
| 55 |
},
|
| 56 |
+
"device": "cpu",
|
| 57 |
"use_amp": false,
|
| 58 |
"use_peft": false,
|
| 59 |
"push_to_hub": true,
|
|
|
|
| 62 |
"tags": null,
|
| 63 |
"license": null,
|
| 64 |
"pretrained_path": null,
|
| 65 |
+
"vjepa_model_id": "facebook/vjepa2-vitg-fpc64-256",
|
| 66 |
+
"trace_horizon": 20,
|
| 67 |
+
"trace_predictor_ckpt": "checkpoints/trace_predictor_final.pth",
|
| 68 |
+
"video_key": "observation.video",
|
| 69 |
+
"language_key": "observation.language",
|
| 70 |
+
"chunk_size": 20,
|
| 71 |
+
"n_action_steps": 20,
|
| 72 |
+
"drop_n_last_frames": 10,
|
| 73 |
"ee_pixel_image_resolution": 256,
|
| 74 |
"normalization_mapping": {
|
| 75 |
"VISUAL": "MEAN_STD",
|
| 76 |
+
"STATE": "MEAN_STD",
|
| 77 |
"ACTION": "MIN_MAX"
|
| 78 |
},
|
|
|
|
| 79 |
"vision_backbone": "resnet18",
|
| 80 |
+
"pretrained_backbone_weights": "ResNet18_Weights.IMAGENET1K_V1",
|
| 81 |
+
"replace_final_stride_with_dilation": false,
|
| 82 |
+
"pre_norm": false,
|
| 83 |
+
"dim_model": 512,
|
| 84 |
+
"n_heads": 8,
|
| 85 |
+
"dim_feedforward": 3200,
|
| 86 |
+
"feedforward_activation": "relu",
|
| 87 |
+
"n_encoder_layers": 4,
|
| 88 |
+
"n_decoder_layers": 1,
|
| 89 |
+
"use_vae": true,
|
| 90 |
+
"latent_dim": 32,
|
| 91 |
+
"n_vae_encoder_layers": 4,
|
| 92 |
+
"temporal_ensemble_coeff": null,
|
| 93 |
+
"dropout": 0.1,
|
| 94 |
+
"kl_weight": 10.0,
|
| 95 |
+
"optimizer_lr": 1e-05,
|
| 96 |
+
"optimizer_weight_decay": 0.0001,
|
| 97 |
+
"optimizer_lr_backbone": 1e-05,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 98 |
"compile_model": false,
|
| 99 |
+
"compile_mode": "reduce-overhead"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 100 |
}
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5d44b9d6ea1c5763964bd4511223f56c0e3f1e8c12dbe45cddc2b99a25a860f9
|
| 3 |
+
size 5248235688
|
train_config.json
CHANGED
|
@@ -81,8 +81,8 @@
|
|
| 81 |
},
|
| 82 |
"env": null,
|
| 83 |
"policy": {
|
| 84 |
-
"type": "
|
| 85 |
-
"n_obs_steps":
|
| 86 |
"input_features": {
|
| 87 |
"observation.state": {
|
| 88 |
"type": "STATE",
|
|
@@ -135,7 +135,7 @@
|
|
| 135 |
]
|
| 136 |
}
|
| 137 |
},
|
| 138 |
-
"device": "
|
| 139 |
"use_amp": false,
|
| 140 |
"use_peft": false,
|
| 141 |
"push_to_hub": true,
|
|
@@ -144,98 +144,79 @@
|
|
| 144 |
"tags": null,
|
| 145 |
"license": null,
|
| 146 |
"pretrained_path": null,
|
| 147 |
-
"
|
| 148 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 149 |
"ee_pixel_image_resolution": 256,
|
| 150 |
"normalization_mapping": {
|
| 151 |
"VISUAL": "MEAN_STD",
|
| 152 |
-
"STATE": "
|
| 153 |
"ACTION": "MIN_MAX"
|
| 154 |
},
|
| 155 |
-
"drop_n_last_frames": 7,
|
| 156 |
"vision_backbone": "resnet18",
|
| 157 |
-
"
|
| 158 |
-
|
| 159 |
-
|
| 160 |
-
|
| 161 |
-
"
|
| 162 |
-
"
|
| 163 |
-
"
|
| 164 |
-
"
|
| 165 |
-
"
|
| 166 |
-
"
|
| 167 |
-
|
| 168 |
-
|
| 169 |
-
|
| 170 |
-
|
| 171 |
-
"
|
| 172 |
-
"
|
| 173 |
-
"
|
| 174 |
-
"
|
| 175 |
-
"noise_scheduler_type": "DDPM",
|
| 176 |
-
"num_train_timesteps": 100,
|
| 177 |
-
"beta_schedule": "squaredcos_cap_v2",
|
| 178 |
-
"beta_start": 0.0001,
|
| 179 |
-
"beta_end": 0.02,
|
| 180 |
-
"prediction_type": "epsilon",
|
| 181 |
-
"clip_sample": true,
|
| 182 |
-
"clip_sample_range": 1.0,
|
| 183 |
-
"num_inference_steps": null,
|
| 184 |
-
"do_mask_loss_for_padding": false,
|
| 185 |
"compile_model": false,
|
| 186 |
-
"compile_mode": "
|
| 187 |
-
"optimizer_lr": 0.0001,
|
| 188 |
-
"optimizer_betas": [
|
| 189 |
-
0.95,
|
| 190 |
-
0.999
|
| 191 |
-
],
|
| 192 |
-
"optimizer_eps": 1e-08,
|
| 193 |
-
"optimizer_weight_decay": 1e-06,
|
| 194 |
-
"scheduler_name": "cosine",
|
| 195 |
-
"scheduler_warmup_steps": 500
|
| 196 |
},
|
| 197 |
-
"output_dir": "outputs/train/2026-02-
|
| 198 |
-
"job_name": "
|
| 199 |
"resume": false,
|
| 200 |
"seed": 0,
|
| 201 |
-
"num_workers":
|
| 202 |
-
"
|
| 203 |
-
"
|
| 204 |
-
"steps": 50000,
|
| 205 |
"eval_freq": 20000,
|
| 206 |
-
"log_freq":
|
| 207 |
"tolerance_s": 0.0001,
|
| 208 |
"save_checkpoint": true,
|
| 209 |
-
"save_freq":
|
| 210 |
"use_policy_training_preset": true,
|
| 211 |
"optimizer": {
|
| 212 |
-
"type": "
|
| 213 |
-
"lr":
|
| 214 |
-
"weight_decay":
|
| 215 |
"grad_clip_norm": 10.0,
|
| 216 |
"betas": [
|
| 217 |
-
0.
|
| 218 |
0.999
|
| 219 |
],
|
| 220 |
"eps": 1e-08
|
| 221 |
},
|
| 222 |
-
"scheduler":
|
| 223 |
-
"type": "diffuser",
|
| 224 |
-
"num_warmup_steps": 500,
|
| 225 |
-
"name": "cosine"
|
| 226 |
-
},
|
| 227 |
"eval": {
|
| 228 |
"n_episodes": 50,
|
| 229 |
"batch_size": 50,
|
| 230 |
"use_async_envs": false
|
| 231 |
},
|
| 232 |
"wandb": {
|
| 233 |
-
"enable":
|
| 234 |
"disable_artifact": false,
|
| 235 |
"project": "lerobot",
|
| 236 |
"entity": null,
|
| 237 |
"notes": null,
|
| 238 |
-
"run_id":
|
| 239 |
"mode": null
|
| 240 |
},
|
| 241 |
"peft": null,
|
|
|
|
| 81 |
},
|
| 82 |
"env": null,
|
| 83 |
"policy": {
|
| 84 |
+
"type": "trace_act",
|
| 85 |
+
"n_obs_steps": 1,
|
| 86 |
"input_features": {
|
| 87 |
"observation.state": {
|
| 88 |
"type": "STATE",
|
|
|
|
| 135 |
]
|
| 136 |
}
|
| 137 |
},
|
| 138 |
+
"device": "cpu",
|
| 139 |
"use_amp": false,
|
| 140 |
"use_peft": false,
|
| 141 |
"push_to_hub": true,
|
|
|
|
| 144 |
"tags": null,
|
| 145 |
"license": null,
|
| 146 |
"pretrained_path": null,
|
| 147 |
+
"vjepa_model_id": "facebook/vjepa2-vitg-fpc64-256",
|
| 148 |
+
"trace_horizon": 20,
|
| 149 |
+
"trace_predictor_ckpt": "checkpoints/trace_predictor_final.pth",
|
| 150 |
+
"video_key": "observation.video",
|
| 151 |
+
"language_key": "observation.language",
|
| 152 |
+
"chunk_size": 20,
|
| 153 |
+
"n_action_steps": 20,
|
| 154 |
+
"drop_n_last_frames": 10,
|
| 155 |
"ee_pixel_image_resolution": 256,
|
| 156 |
"normalization_mapping": {
|
| 157 |
"VISUAL": "MEAN_STD",
|
| 158 |
+
"STATE": "MEAN_STD",
|
| 159 |
"ACTION": "MIN_MAX"
|
| 160 |
},
|
|
|
|
| 161 |
"vision_backbone": "resnet18",
|
| 162 |
+
"pretrained_backbone_weights": "ResNet18_Weights.IMAGENET1K_V1",
|
| 163 |
+
"replace_final_stride_with_dilation": false,
|
| 164 |
+
"pre_norm": false,
|
| 165 |
+
"dim_model": 512,
|
| 166 |
+
"n_heads": 8,
|
| 167 |
+
"dim_feedforward": 3200,
|
| 168 |
+
"feedforward_activation": "relu",
|
| 169 |
+
"n_encoder_layers": 4,
|
| 170 |
+
"n_decoder_layers": 1,
|
| 171 |
+
"use_vae": true,
|
| 172 |
+
"latent_dim": 32,
|
| 173 |
+
"n_vae_encoder_layers": 4,
|
| 174 |
+
"temporal_ensemble_coeff": null,
|
| 175 |
+
"dropout": 0.1,
|
| 176 |
+
"kl_weight": 10.0,
|
| 177 |
+
"optimizer_lr": 1e-05,
|
| 178 |
+
"optimizer_weight_decay": 0.0001,
|
| 179 |
+
"optimizer_lr_backbone": 1e-05,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 180 |
"compile_model": false,
|
| 181 |
+
"compile_mode": "reduce-overhead"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 182 |
},
|
| 183 |
+
"output_dir": "outputs/train/2026-02-17/20-33-57_trace_act",
|
| 184 |
+
"job_name": "trace_act",
|
| 185 |
"resume": false,
|
| 186 |
"seed": 0,
|
| 187 |
+
"num_workers": 0,
|
| 188 |
+
"batch_size": 2,
|
| 189 |
+
"steps": 2,
|
|
|
|
| 190 |
"eval_freq": 20000,
|
| 191 |
+
"log_freq": 1,
|
| 192 |
"tolerance_s": 0.0001,
|
| 193 |
"save_checkpoint": true,
|
| 194 |
+
"save_freq": 10,
|
| 195 |
"use_policy_training_preset": true,
|
| 196 |
"optimizer": {
|
| 197 |
+
"type": "adamw",
|
| 198 |
+
"lr": 1e-05,
|
| 199 |
+
"weight_decay": 0.0001,
|
| 200 |
"grad_clip_norm": 10.0,
|
| 201 |
"betas": [
|
| 202 |
+
0.9,
|
| 203 |
0.999
|
| 204 |
],
|
| 205 |
"eps": 1e-08
|
| 206 |
},
|
| 207 |
+
"scheduler": null,
|
|
|
|
|
|
|
|
|
|
|
|
|
| 208 |
"eval": {
|
| 209 |
"n_episodes": 50,
|
| 210 |
"batch_size": 50,
|
| 211 |
"use_async_envs": false
|
| 212 |
},
|
| 213 |
"wandb": {
|
| 214 |
+
"enable": false,
|
| 215 |
"disable_artifact": false,
|
| 216 |
"project": "lerobot",
|
| 217 |
"entity": null,
|
| 218 |
"notes": null,
|
| 219 |
+
"run_id": null,
|
| 220 |
"mode": null
|
| 221 |
},
|
| 222 |
"peft": null,
|